Eplore ML in Sklearn Wrappers

Usage

Easily explore parameters and models of sklearn, XGBoost, etc. using a group of helper functions. Run a simulation using the following steps

1. Load Data: i.e. pd.read_csv(...)
2. Define Regressors and Parameter Grids: models = [Regressor("name", RegressorObject, [{'param1':value1},{'param2':value2}]) (see src/DefineModels.py)
3. Define DataSettings: `ds = DataSetting(y=data.y, x=data[regressors], models=models.copy(), loss_function=rmse, k=5)
4. Run evaluate_all()
5. Plot validation or losses: i.e. `ds.plot_model_validation_curves(path=output_path)

Advantage

A set of regressors and parameters can easily be extended and experiments can be conducted in a controlled manner. The same settings can be applied to different datasets easily. Easy plots of progress with respect to different variables.

Example Validation Curve	Example Learning Curves

Disadvantage

Parallelisation not implemented. Breaking out of the predefined framework requires reworking the imported classes.